Abstract
Background: The evolution of the primate brain has been characterized by the reorganization of key structures and circuits underlying derived specializations in sensory systems, as well as social behavior and cognition. Among these, expansion and elaboration of the prefrontal cortex has been accompanied by alterations to the connectivity and organization of subcortical structures, including the striatum and amygdala, underlying advanced aspects of executive function, inhibitory behavioral control, and socioemotional cognition seen in our lineages. At the cellular level, the primate brain has further seen an increase in the diversity and number of inhibitory GABAergic interneurons. A prevailing hypothesis holds that disruptions in the balance of excitatory to inhibitory activity in the brain underlies the pathophysiology of many neurodevelopmental and psychiatric disorders. Summary: This review highlights the evolution of inhibitory brain systems and circuits and suggests that recent evolutionary modifications to GABAergic circuitry may provide the substrate for vulnerability to aberrant neurodevelopment. We further discuss how modifications to primate and human social organization and life history may shape brain development in ways that contribute to neurodivergence and the origins of neurodevelopmental disorders. Key Messages: Many brain systems have seen functional reorganization in the mammalian, primate, and human brain. Alterations to inhibitory circuitry in frontostriatal and frontoamygdalar systems support changes in social behavior and cognition. Increased complexity of inhibitory systems may underlie vulnerabilities to neurodevelopmental and psychiatric disorders, including autism and schizophrenia. Changes observed in Williams syndrome may further elucidate the mechanisms by which alterations in inhibitory systems lead to changes in behavior and cognition. Developmental processes, including altered neuroimmune function and age-related vulnerability of inhibitory cells and synapses, may lead to worsening symptomatology in neurodevelopmental and psychiatric disorders.
Introduction
The evolution of the human brain has been marked by a threefold increase in volume in our lineage since divergence from our last common ancestor shared with the living great apes. Much attention has been paid to the novelty and reorganization of the prefrontal cortex (PFC) in primate and human lineages and the emergence and differential expansion of association cortices and their unique computational properties. Alongside these changes, subcortical structures sharing connections with the PFC have additionally seen substantial reorganization of their extrinsic and intrinsic connections, as well as internal circuitry. Among these, functional loops within the basal ganglia and amygdala subserve highly complex adaptations related to inhibitory behavioral control and social cognition, which will be the focus of this review. An overarching theme of this review is inhibition, which represents multiple levels of analysis, first from a broader systems view focusing on the contributions of the PFC and closely connected structures including the basal ganglia and amygdala and next to the level of a more fine-grained approach encompassing the contributions of GABAergic systems in the brain to circuit level function – and ultimately dysfunction in the case of neurodevelopmental and psychiatric disorders, particularly those affecting social behavior and cognition. As a means of breaking down these levels, we will focus on some known derived features in the circuitry of the human PFC, striatum, and amygdala specifically and highlight the changes in these regions that arise as a function of neurodevelopmental disorders and their distinctive pathophysiology, with a particular focus on Williams syndrome (WS).
Cortex and Inhibitory Systems
A hallmark feature of mammalian brain evolution, the six-layered neocortex, was likely present in the last common ancestor of living mammals including primates, some 200 million years ago [1]. Systematic variation in the structure and organization of cortical layers in part determines their function across species and reflects their ontogeny and evolution [2, 3]. Neural networks broadly consist of glutamatergic excitatory neurons and inhibitory local interneurons. This arrangement is phylogenetically quite ancient, and both glutamate and gamma-aminobutyric acid (GABA) serve as key neurotransmitters in animals with the most rudimentary of nervous systems, including insects, roundworms, and platyhelminths [4]. GABA was likely present in prokaryotes but did not function as a neurotransmitter until the rise of bilaterians. In ctenophores, a lack of pharmacological effects on behavior and motor systems suggest that the presence of GABA is a passive by-product of glutamate metabolism [5]. In non-bilateral organisms, glycine functions as an inhibitory molecule [6], though, in ctenophores, glycine can also stimulate ionotropic glutamate receptors [7]. Migration of excitatory projection neurons in the neocortex from pallial neural progenitors has been thought to underlie the evolution of the mammalian neocortex; however, evolutionary changes in the migratory ability of interneurons derived from the medial ganglionic eminence in the ancestor of mammals seems to have been involved in the establishment of the neocortex by supplying inhibitory cells to cortical systems [8, 9]. Single-cell transcriptomic analyses of reptilian brains have further revealed that inhibitory interneuron classes present in mammals were also likely present in the common ancestor of all amniotes [10].
GABAergic inhibitory interneurons comprise 15% or less of the cortical neuron population in rodents and other non-primate mammalian species, whereas they may constitute more than 20% of neurons within the primate neocortex [11]. The different origins and life history of GABAergic interneuron subpopulations likely relate to differences in patterning among taxonomic groups within the neocortex [12]. Specifically, the embryonic origin and migration of GABAergic cells has been demonstrated to differ between rodents and primates, with additional sites of neurogenesis in the lateral ventricular neuroepithelium in primates [13]. Timing of the emergence and maturation of these cells shows interesting patterns of variability across taxa and includes a more protracted period of neuronal production shown in primate lineages [14]. The dual origin of cortical interneurons likely contributes to greater diversity in interneuron subpopulations found in the primate neocortex [15, 16]. In the primate temporal lobe, the amygdala, hippocampus, and neocortical subventricular zone are ventrally positioned, and receive migrating neurons from the inferior ganglionic eminence, in a pattern not observed in rodents [17]. The subpial granular layer found in primates and humans additionally provides a novel source of migrating interneurons to layer I of the neocortex [18, 19].
Inhibitory interneurons are typically classified into overlapping subpopulations based on their immunoreactivity for proteins and neuropeptides including vasoactive intestinal peptide (VIP), neuropeptide Y, cholecystokinin (CCK), somatostatin (SST) and choline acetyltransferase, and the calcium binding proteins calbindin (CB), calretinin (CR), and parvalbumin (PV). Interneurons immunoreactive for the calcium binding proteins CB, CR, and PV colocalize with GABA and account for more than 90% of all cortical interneurons, with limited overlap among the separate populations [20‒22], and will be the primary focus of this review. Different classes of interneurons interact with pyramidal neurons to modulate cortical circuit processing. While CB and CR interneurons are involved mostly in intracolumnar communication, PV interneurons regulate transcolumnar signaling. PV interneurons in the PFC synchronize the activity of pyramidal cells through perisomatic and axo-axonic inhibition. Each class of inhibitory neurons has a unique pattern of laminar distribution and innervates specific segments of neighboring neurons. Studies of human fetal brain tissue further suggest that more complex neuronal progenitor populations give rise to neocortical interneurons relative to rodents, particularly CR-positive interneurons [19]. CR inhibitory neurons are found mostly in superficial layers I-IIa, whereas CB neurons are concentrated mostly in layers II-IIIa. PV labels the basket and chandelier neurons, which are found predominantly in the middle-deep layers V-VI (DeFelipe [20]) and innervate perisomatic elements of nearby neurons. Interneuron subtypes also vary in density by region. For example, in anterior cingulate and agranular orbitofrontal PFC areas, CB neurons are more densely distributed as compared to PV neurons [23], whereas in lateral eulaminate areas, they are more balanced.
Modifications to the structure, function, and distribution of interneuron types within and across cortical columns support differences in cognitive abilities between taxonomic groups [24, 25]. For example, CB double bouquet cells (DBCs) are found in primate cortex, and not in rodents, but are largely restricted to visual cortex in carnivores [26] suggesting specialization of minicolumnar inhibition unique to primates through a unique pattern of axonal connectivity. Since the last common ancestor of Euarchontoglires and laurasiatherians likely did not share DBCs, it is more plausible that the existence of DBCs in primates and carnivores represents the convergent evolution of morphologically similar cell type that is not homologous.
PFC: Primate-Specific Modifications
The precise definition of the PFC, and its presence in other species, particularly rodents, has been the subject of debate (see [27, 28] for discussion) but is universally recognized as an area of significant novelty and expansion in primate evolution. Agranular territories of the ventromedial PFC and anterior cingulate cortex including Brodmann areas (BAs) 24 and 32 of the primate brain more closely resemble PFC of rodents, and serve as a critical interface for integration of limbic and sensory information. Eulaminate cortices, defined by the presence of a clear layer IV and connectivity with the mediodorsal nucleus of the thalamus, are found in PFC territories of primates, with unique functional architectonics that subserve highly complex multimodal processing. Select regions of PFC association cortex are disproportionately enlarged in humans as compared to chimpanzees and bonobos, and expansion and elaboration of PFC regions has been associated with adaptations in social behavior and cognition that have defined primate and human evolution (see Teffer and Semendeferi [29], for review; Wei et al. [30]). These changes subserve unique cognitive faculties and include territories involved in executive function, such as BAs 9 and 46 of the dorsolateral PFC, language and speech production (BA 44/45), and socioemotional processing (BAs 47/12, 10, 11, and 13). While interneuron subtypes display broadly conserved patterns of distribution in the PFC and the expression of associated genes among primates [31], it is likely that alterations to the connectivity and function of cortical assemblages resulting from the expansion of association cortices contribute to derived human specializations in cognition and behavior. Ma and colleagues [32] performed single-nucleus RNA-seq analyses in the dorsolateral PFC of humans, chimpanzees, macaques and marmosets and found broad conservation of major transcriptomically defined cell types, including excitatory and inhibitory interneurons, astrocytes, and oligodendrocytes, with notable differences in expression profiles. Among these, inhibitory neurons showed changes in genes coding for neuropeptides and carbonic anhydrases critical for cell-cell communication and neuronal signaling, suggesting reorganization of synaptic communication in inhibitory neurons across primate lineages.
Neurotransmitter-specific pathways from the basal forebrain and brainstem interact with neuronal populations in the PFC, including cholinergic, serotonergic, and dopaminergic pathways. These neurotransmitter systems are highly involved in the modulation of behavioral inhibition, as well as behavioral flexibility, learning, attention, and reinforcement of behavior, and their altered distribution in apes and humans may reflect derived specializations in these cognitive faculties. Specifically, humans and apes show an increased density of serotonergic fibers as compared to macaques [33]. Cholinergic fiber density does not appear to differ between primate taxa, but the presence of fiber clusters indicative of cortical plasticity events was noted in humans and chimpanzees but not in macaques [34]. Similarly, coils of serotonergic axons were also observed in humans and chimpanzees but not macaques [33]. Unique patterns of laminar distribution show specializations of dopaminergic systems across primate taxa in PFC regions BA 9 and 32, with humans displaying increased dopaminergic afferents in layers III and V/VI [35], and the presence of tyrosine hydroxylase-positive neurons, which were found in humans as well as monkey species examined and in siamangs but were notably absent in all nonhuman great ape taxa examined [36]. In light of these human-specific changes in dopaminergic interneuron distribution, Sousa and colleagues [37] further examined cell-specific expression patterns in the primate brain, confirming the absence of these rare, subpallial-derived neurons in the nonhuman African ape neocortex, as well as enrichment of genes associated with dopamine biosynthesis in the human striatum. Further work in the dorsolateral PFC [38] suggested a human-specific switching of expression between SST and tyrosine hydroxylase among these neurons, as well as species-specific differences in dopaminergic synthesis, suggesting key modifications to dopaminergic signaling pathways in the human lineage.
Subcortical Structures: The Striatum and Amygdala in Primate Evolution
The primate PFC shares extensive connectivity with subcortical territories, including the basal ganglia and the amygdala, which subserve diverse facets of cognitive and emotional processing, as well as modulating reward seeking and risk avoidance behavior. These circuits provide reciprocal connections that integrate behavior, emotion, and action and provide the substrates for the inhibitory control of behavior – an important facet of social life.
Examining allometric relationships between structures may be informative for understanding their functional role in phylogenetically relevant adaptations [39‒41], though intraspecific variation and small sample size may present challenges in drawing meaningful conclusions regarding brain structures [42]. To examine the role of neural structures involved in socioemotional processing and cognition, we performed an evolutionary analysis of new and previously published data on primate limbic structures using a large sample of previously published and newly collected data from a diverse array of primate taxa [43]. Compared with other hominoids, the volumes of the hippocampus, the lateral nucleus of the amygdala, and the orbitofrontal cortex were greater in humans than predicted for an ape of human hemisphere volume, while the medial and dorsal frontal cortex was significantly smaller than expected. Compared with other anthropoids, only human values for the striatum fell significantly below predicted values. Overall, these data support the observation that regions associated with emotional processing are not de-emphasized in primate lineages, as well as hinting at reorganization of neural circuitry and connectivity underlying these changes.
Allometric comparisons raise important questions regarding the impact of volumetric differences on these structures, their connectivity, and differences in internal microstructure between taxa. Application of modern methods, including single-cell multi-omics and spatial transcriptomics, has significantly expanded the utility of cross-species studies and allowed for greater understanding of phylogenetic variation in cell types and their origin, diversity, and distribution among primate species [44‒47]. While most primate-centered studies have focused on the highly derived cerebral cortex, changes in cellular composition and gene expression have also been noted in subcortical structures, including the striatum, amygdala, and cerebellum [46, 48].
Sophisticated social behavior and emotional processing involve precise coordination between the cortex and many additional subcortical territories, including broader interactions between lateral, orbital, and medial prefrontal as well as temporal and parietal cortices, the hippocampus and entorhinal cortex, basal forebrain structures, thalamocortical systems, the cerebellum, and many additional structures and systems that subserve complex communication and interaction. While definitions of “social” brain regions vary widely, most acknowledge the relationship between reward systems and systems for behavioral control as critical for social interaction. As such, we have focused our discussion on two prefrontal-subcortical networks – rooted in connections between the PFC and the striatum and amygdala, specifically – as parallel networks important for understanding inhibitory control of behavior.
Cortical-Striatal Systems for Behavioral Control
The structure and function of the basal ganglia have been described as highly conserved, with recognizable basic components arguably present in lamprey [49] and found in the last common ancestor of vertebrates more than 500 million years ago [50]. While tetrapod vertebrates share a common pattern of basal ganglia structural integration, a trend toward increased involvement of the cortex in the integration of multimodal sensory information from the thalamus has followed the expansion and elaboration of the cortex in mammals [51]. The striatum is composed of the dorsal striatum, including the caudate and putamen, and the ventral striatum, inclusive of the nucleus accumbens region. Importantly, the striatum serves as a principal input for glutamatergic and dopaminergic inputs from the cortex and subcortical territories and is a site of convergence for the orchestration of motor behavior and cognition. In both rodents and primates, the striatum is characterized by the topographical organization of inputs from the frontal cortex (Fig. 1). Parallel and integrated networks subserve motor, cognitive, and reward-related functions. Research has shown a high degree of homology in these inputs from the medial PFC in rodent and primate striatum [52‒54]. Just as the primate PFC is characterized by the emergence of complex granular prefrontal regions, and the human brain is characterized by further elaboration of those territories, the diversity of inputs to striatal territories in the primate and human brain includes an enrichment of projections from the orbitofrontal and dorsolateral PFC [55‒57]. These overlapping projections support enrichment of executive function and alterations to associated circuitry and contribute to inhibitory behavioral control and reward processing.
Similarly to the cortex, the primate striatum is characterized by a greater percentage of interneurons of diverse subtypes as compared to the rodent [58, 59], though medium spiny projection neurons still significantly predominate. GABAergic interneurons have been extensively characterized for their neurochemical and electrophysiological properties and contribute to a diverse and heterogeneous local architecture [60, 61]. Cholinergic interneurons, which are excitatory, comprise an additional 1% of striatal neurons in the human striatum [62, 63] and share extensive connections with cortical and subcortical projections with vast networks of axonal varicosities [64, 65]. They receive direct projections from the substantia nigra, thalamus, and cortex and play a significant role in modulating inputs to local inhibitory interneurons and medium spiny projection neurons [66, 67]. These cells may have a particularly significant role in modulating appropriate social behavior [68‒70] as shown in animal models.
Transcriptomic analyses have identified a unique population of striatal interneurons, accounting for some 30% of total striatal interneurons in the primate. These recently characterized TAC3 interneurons derive from the medial geniculate eminence and are defined by unique transcriptional and developmental patterns [71, 72]. Transcriptomic analyses of developing striatal interneurons in the primate brain [72] have further shown that olfactory bulb precursor cells are systematically redirected to the striatum and cortex. The authors suggest an evolutionary model wherein conserved classes of neurons supplying the smaller primate olfactory bulb are reused or re-routed in development in the striatum and cortex. This kind of “recycling” as exaptation and developmental reprogramming may provide a plausible substrate for systemic expansion and reorganization of neural circuitry.
In the human striatum, we see further examples of derived molecular and cellular differences in brain organization. Analyses of postmortem tissue from human, chimpanzee, and macaque brains [73] have found the highest number of differentially expressed genes was in the striatum, characterized by a human-specific up-regulation of dopamine biosynthesis and signaling genes. Humans were further shown to have a greater number of dopaminergic interneurons in both the dorsal caudate nucleus and putamen when compared with a broader sample of 9 other taxa of nonhuman primates. The human striatum further displays a unique neurochemical profile characterized by elevated striatal dopamine, serotonin, and neuropeptide Y, as well as reduced acetylcholine, which has been hypothesized to amplify sensitivity to social cues and a shift toward affiliative and pro-social behavior in recent human evolution [74].
PFC-Amygdala Connections: Pathways for Emotional Processing
The primate amygdala shares extensive connections with the PFC (Fig. 2) that subserve important contributions to phylogenetically specific behavioral and cognitive adaptations (see [75] for review). In mammals, the heterogeneous amygdala nuclei are broadly grouped into three major subdivisions. The deep nuclei include the basal, accessory basal, and lateral nuclei that are densely interconnected with the PFC. The superficial nuclei consist of the periamygdaloid cortex, cortical nuclei, and medial nucleus. The central nucleus and anterior amygdaloid area have major output to the autonomic regions of the hypothalamus and brainstem.
The deep, basolateral nuclei and superficial, cortical nuclei are largely derived from the pallium, which gives rise to glutamatergic cells of the cortical mantle in development. The medial and central nuclei derive from subpallial structures, which supply progenitors for GABAergic neurons [76]. As compared to rodents, primates display enlarged basal, accessory basal, and lateral nuclei [77, 78], attributed to the expansion of richly interconnected frontal and parietal cortices. The medial and cortical amygdalae, which receive extensive projections from the olfactory and vomeronasal system, are diminished in relative size in primates in comparison to the basal, lateral, and accessory basal nuclei, a shift from the pattern observed in more olfactory-reliant rodents. Although very similar in cytoarchitecture and connectivity, the amygdala has seen a reorganization in cell number in human evolution [79]. Among hominoids, the human lateral nucleus contains the highest number of neurons in the amygdala, whereas in nonhuman apes, the basal nucleus contained the highest number of neurons. Additionally, the human lateral nucleus contains 59% more neurons than predicted by allometric regressions on nonhuman primate data, underscoring its emphasis over the course of human evolution.
Inhibitory circuits of the basolateral amygdala orchestrate emotional responses to stimuli, guided by populations of GABAergic interneurons [80]. Inhibitory interneuron subtypes comprise about 20% of neurons in the basolateral division [81‒83], of which about half express CB and some 20% are CR positive. CB-expressing interneurons can be further divided into those that express PV, SST, or cholecystokinin. Networks of PV cells may act to modulate basolateral activity through strong connections with projection neurons and to synchronize the firing of principal, projecting cells. Qualitatively, PV interneuron density is similar across rat, monkey, and human, with the highest density within the lateral nucleus and sparse in more medial nuclei [81]. The lateral nucleus contains the highest density and the basal, accessory basal, and central nuclei show a decreasing density (Pantazopoulos et al. [84], Sorvari et al. [85], García-Amado and Prensa [86]). Although the central nucleus is densely populated with GABAergic interneurons in macaques [83], PV+ interneurons are not the most prevalent population in this region for humans and are quite sparsely distributed (García-Amado and Prensa [87]). The central nucleus consists entirely of GABAergic projection neurons, which share similarities in developmental origin and physiological properties with medium spiny neurons in the striatum [83, 88‒90]. Intercalated cell masses serve as an interface between the basolateral and central divisions, providing feed-forward inhibition to the central nucleus, modulating fear conditioning [91‒94].
The amygdala is also heavily innervated by serotonergic projections, which contribute significantly to the modulation of local activity and in part act to regulate inhibitory control of amygdala function by the PFC [95]. Although all 5-HT receptor subtypes associated with serotonergic processing have been documented in the amygdala, most (∼65–70%) GABA-immunoreactive neurons in the basolateral amygdala exhibit immunoreactivity for 5HT2A receptors [96]. Previous research [97] has shown differences in the distribution of serotonin transporter-immunoreactive (SERT-ir) axons between bonobos and chimpanzees, such that the amygdala of bonobos contains more than twice the density of serotonergic axons than chimpanzees, with the most pronounced differences in the basal and central nuclei. These findings suggest that variation in serotonergic innervation of the amygdala [98] may contribute to mediating the remarkable differences in social behavior exhibited by bonobos and chimpanzees, including reduced risk preference and emotional reactivity, and increased sociosexual behavior in bonobos, as well as greater aggression in chimpanzees. The human pattern of serotonergic axon distribution in the amygdala additionally complements the redistribution of neurons in the amygdala in human evolution. Humans displayed a unique pattern of serotonergic innervation of the amygdala, and SERT-ir axon density was significantly greater in the central nucleus compared to the lateral nucleus. SERT-ir axon density was significantly greater in humans compared to chimpanzees in the basal, accessory basal, and central nuclei (Lew et al. [234]). SERT-ir axon density was greater in humans compared to bonobos in the accessory basal and central nuclei. These findings suggest that differential serotonergic modulation of cognitive and autonomic pathways in the amygdala in humans, bonobos, and chimpanzees may contribute to species-level differences in social behavior.
Dysfunction of Inhibitory Circuitry in Neurodevelopmental Disorders
Variation in the distribution of GABAergic interneurons has been shown in neurodevelopmental and psychiatric disorders with diverse effects on behavioral and cognitive symptomatology (Table 1). For example, the density of PV interneurons is significantly reduced in ASD compared with controls in the dorsolateral PFC [99], which is involved in executive functions including working memory and attention. The number of CB and CR interneurons did not differ in these regions. Reduction in PV interneurons is hypothesized to disrupt the balance between excitation and inhibition and alter gamma wave oscillations in the cerebral cortex of ASD subjects. The decrease in numbers of PV interneurons was further shown to more strongly affect chandelier cells [100], as compared to basket cells. By contrast, chandelier cells are axo-axonic and synapse onto the axon initial segment of excitatory pyramidal cells. Importantly, each chandelier cell can innervate hundreds of different pyramidal cells to synchronize activity across a wide-reaching cellular assemblage. Thus, chandelier cells can directly influence cortical circuit function, and the loss of a single chandelier cell may have widespread effects on circuitry and the excitation/inhibition (E/I) balance. The loss of chandelier cells and cartridges may have implications for the course of symptomatology including seizure activity in the brain in ASD and other neurodevelopmental disorders, via the imbalance of excitatory to inhibitory neuronal circuitry [101, 102], although it is worth noting that seizure activity in ASD may represent diverse cellular mechanisms involving many interneuronal subtypes [103, 104]. Importantly, the loss of chandelier cell cartridges may correlate with severity of intellectual impairment [105].
Brain region . | Disorder . | Cell population . | Key finding . | Citation . |
---|---|---|---|---|
PFC | ASD | PV+ chandelier cells | ↓chandelier cells | [99] |
↓chandelier cell cartridges | [100] | |||
Schizophrenia | PV+ basket cells | ↓perineuronal net density in late adolescence/early adulthood | [106] | |
PV+ interneurons | ↓PV+ neuron density, reduced mRNA | [107] (meta-analysis) | ||
Bipolar disorder | PV+ basket cells | No difference in perineuronal nets | [106] | |
Striatum | ASD | CR+ | ↓ in caudate | [108] |
Schizophrenia | Ch+ interneurons | ↓density in medial caudate, nucleus accumbens | [109, 110] | |
NOS+ interneurons | ↓density in putamen | [111] | ||
WS | Ch+ interneurons | ↓density in medial caudate, nucleus accumbens | [112] | |
PV+ interneurons | No differences | |||
Tourette syndrome | PV+ interneurons | ↓density in caudate, putamen | [113, 114] | |
Ch+ interneurons | ↓density in caudate, putamen | [113] | ||
Amygdala | ASD | |||
Schizophrenia | PV+ interneurons | ↓density in lateral, basal | [115] | |
SST+ interneurons | ↓density in lateral nucleus | [116] | ||
WS | PV+ interneurons | No significant difference | [117] | |
Bipolar disorder | SST+ interneurons | ↓density in lateral nucleus | [116] |
Brain region . | Disorder . | Cell population . | Key finding . | Citation . |
---|---|---|---|---|
PFC | ASD | PV+ chandelier cells | ↓chandelier cells | [99] |
↓chandelier cell cartridges | [100] | |||
Schizophrenia | PV+ basket cells | ↓perineuronal net density in late adolescence/early adulthood | [106] | |
PV+ interneurons | ↓PV+ neuron density, reduced mRNA | [107] (meta-analysis) | ||
Bipolar disorder | PV+ basket cells | No difference in perineuronal nets | [106] | |
Striatum | ASD | CR+ | ↓ in caudate | [108] |
Schizophrenia | Ch+ interneurons | ↓density in medial caudate, nucleus accumbens | [109, 110] | |
NOS+ interneurons | ↓density in putamen | [111] | ||
WS | Ch+ interneurons | ↓density in medial caudate, nucleus accumbens | [112] | |
PV+ interneurons | No differences | |||
Tourette syndrome | PV+ interneurons | ↓density in caudate, putamen | [113, 114] | |
Ch+ interneurons | ↓density in caudate, putamen | [113] | ||
Amygdala | ASD | |||
Schizophrenia | PV+ interneurons | ↓density in lateral, basal | [115] | |
SST+ interneurons | ↓density in lateral nucleus | [116] | ||
WS | PV+ interneurons | No significant difference | [117] | |
Bipolar disorder | SST+ interneurons | ↓density in lateral nucleus | [116] |
In schizophrenia, the picture regarding potential GABAergic interneuron loss, reduced mRNA expression, and circuit level alterations in PFC is complex (Dienel and Lewis [118], Stan and Lewis [119], Curley et al. [120], Lewis et al. [121], Lewis et al. [122]). Although total neuron number is not decreased in the PFC in schizophrenia [123], the density of GAD67 mRNA expression is 25–35% lower across cortical layers I–V [124, 125]. While early studies reported a lower density of PV neurons [126, 127], these findings may reflect methodological confounds due to low overall PV levels (Stan and Lewis [119]). It is likely that a subset of PV+ cells – specifically basket cells – are differentially affected in schizophrenia and psychosis [106, 128, 129], in contrast to chandelier cell alterations in ASD, particularly with regard to disruptions to perineuronal nets [128]. Perineuronal net disruption may reflect neuroimmune dysregulation in schizophrenia, as well as in ASD [130], where perineuronal net disruption has also been shown in the globus pallidus of the basal ganglia [131]. Taken together, perineuronal net disruption, mediated by altered neuroimmune function, may reflect shared pathways for differing symptomatology in ASD and schizophrenia, where altered neuroimmune function is a common factor [132, 133]. In concert with genetic and other environmental factors, maternal immune activation may provide a plausible mechanism for shared neuroimmune disruption across disorders [134], which we will discuss later in this review.
WS and the Disinhibition of Social Behavior
WS is a rare neurodevelopmental disorder caused by a hemizygous deletion in chromosomal region 7q11.23, resulting in a distinctive hypersocial phenotype characterized by disinhibited social approach, as well as preserved or even unusually expressive language function [135‒137]. Several genes in the WS deletion, including GTF2IRD1, CYCLN2, LIMK, FZD9, and DNAJC30, have been implicated in social cognition, emotional and fear conditioning, neuronal development, and social anxiety-like behaviors [138‒145]. The phenotype presents a stark contrast to ASD, which comprises a diverse spectrum of altered social cognition and behavior as compared to neurotypical (NT) individuals principally characterized by deficits in social behavior, language, and restricted interests. Strikingly, a subset of cases of ASD have been shown to involve a duplication of the WS deleted region [146, 147], presenting a spectrum of variation defined by hypersociality and relatively preserved language function in WS at one extreme and pronounced speech and language deficits and reduced social behavior in duplication 7q11.23 syndrome [148] on the other. Notably, this region has been analyzed across a range of nonhuman primate species and has shown evidence for recent and rapid evolution at this locus, indicated by high numbers of segmental duplications [149]. Alu sequence transposable elements located at the edges of the deleted region may have served as a source of very important interspecific variation in the anthropoid primate lineage that may predispose this region of the genome to genetic disruption [150]. Thus, the region implicated in WS and duplication 7q11.23 syndrome may represent a “hot spot” for recent genomic change and a potential site of vulnerability in the etiology of neurodevelopmental disorders.
Through our ongoing research, we are beginning to understand the neuroanatomical correlates of the distinct hypersocial phenotype observed in WS. Specific deficits in behavioral response inhibition point to alterations to frontostriatal circuitry [151], which include volumetric reductions in orbitofrontal cortex and striatum [152]. At the cellular level, neuronal density is reduced in the PFC, particularly in layers V/VI [153, 154], which is complemented by increased glial density in BA 25 [154] and altered dendritic morphology [155] as compared to NT controls.
In the striatum, we found no differences in total neuron density between WS and controls but an increase in the density of glial cells, and particularly oligodendrocytes, in the medial caudate nucleus [156], with implications for reduced frontostriatal activation and altered connectivity. We additionally examined the distribution of inhibitory PV and excitatory cholinergic interneurons to better understand the neural circuitry underlying altered inhibitory control and social behavior [112]. We found no difference in the distribution of PV interneurons but a significant reduction in cholinergic interneuron density in the medial caudate and nucleus accumbens regions – areas intimately linked with PFC-striatal connections supporting the processing of social and reward stimuli. Reduced cholinergic interneuron numbers have also been found in similar territories in schizophrenia [109]. In animal models, the selective ablation of striatal cholinergic interneurons was found to induce a behavioral phenotype characterized by aberrant social approach behavior [68], suggesting a plausible mechanism for altered social behavior in both disorders. Of note, the specific location of reduced striatal cholinergic and PV interneuron density in motor territories of the putamen in Tourette’s syndrome [113] likely underlies disinhibited vocal and motor tics. Thus, the pattern of behavioral alterations supported by location of cholinergic interneuron depletion or dysfunction is likely highly location dependent and maps onto the topographical organization of motor and cognitive loops in frontostriatal circuitry.
Given the critical role of the amygdala in social behavior, WS’s unique hypersocial drive, and the role of the inhibitory interneuron system in modulating neuronal circuitry, we targeted interneurons of the amygdala of WS, a potential site of dysfunction. Utilizing immunohistochemical and unbiased stereological techniques, PV+ interneuron density was measured in four nuclei of the amygdala in postmortem tissue of four WS and age-, sex-, and hemisphere-matched control pairs. We hypothesized that the seemingly uninhibited hypersocial WS phenotype may be the result of altered PV+ inhibitory interneuron density with the amygdala reflected in disinhibited social approach, as seen in WS. No significant differences in density or total number were detected in this pilot study (Fig. 3), though the pattern of PV+ interneuron distribution followed the pattern in NT individuals in the present study as well as previously demonstrated (García-Amado and Prensa [87]).
Comparative studies of brain organization in closely related neurodevelopmental disorders provide a powerful opportunity to explore changes in brain circuitry using standardized methods. In the amygdala in WS, the pattern of neuron distribution complements the contrasting sociocognitive phenotype: while the lateral nucleus displays increased numbers of neurons in WS as compared to NT individuals [157], the pattern is reversed in ASD, where a decrease in neuron numbers in the lateral nucleus was found [158]. We additionally compared serotonergic innervation in the amygdala in autism and in WS with the NT human sample, and found opposing directions of change in serotonergic innervation in the two disorders, with ASD displaying an increase in serotonergic axons compared to NT controls and WS displaying a decrease as compared to NT subjects [159]. Differential alterations to cellular distribution and serotonergic innervation of the amygdala may thus contribute to differences in the sociobehavioral phenotype in WS and ASD.
Evolutionary and Life History Factors Supporting Increased Vulnerability to Neurodivergence
A tradeoff in reliance on visual systems over olfactory and other modalities has been a defining characteristic of primate evolution. In anthropoid primates, specializations to the temporal visual cortex have involved the rise of inputs to ventrolateral and granular orbital PFC areas, integrating visual information with decision-making systems [160]. For humans and nonhuman primates, the increased complexity of the social milieu is also associated with the expansion of the brain and particularly the neocortex [161‒164]. The PFC subserves advanced cognitive functions including the integration of information from multiple sensory modalities, directing and maintaining attention, working memory, and the coordination of goal-directed behaviors – all essential facets of primate social life. In humans, volumes of the orbitofrontal [165] and ventromedial [166] PFC regions have been shown to reflect social network size among individuals.
Among primates, humans represent an extreme of a trend toward increased group size and social complexity. This presents additional challenges – including greater risk for exposure to infectious disease. Indeed, a positive association between pathogen-rich environments and high social transmission of behavior and social learning has been demonstrated [167], though some of the risk associated with increased social transmission of behavior and culture may be mitigated by social subgrouping and fission-fusion dynamics [168]. Nevertheless, increased risk of exposure to infectious disease not only presents greater risk of associated morbidity and mortality to the individual, but infection during pregnancy may present unique challenges to offspring neurodevelopment.
Epidemiological studies have suggested that maternal exposure to viral or bacterial infections during pregnancy may increase the risk of offspring neurodevelopmental disorders, including both schizophrenia and ASD [169‒188]. The precise mechanisms underlying altered immune function at the fetal-maternal interface and potential changes in brain development are not clearly understood, and considerations of biological variables including timing of exposure to infection, severity of maternal immune response, offspring genotype and physiology, and offspring sex contribute to controversy in identifying maternal immune activation itself as a causative variable in altered neurodevelopment [189]. Nevertheless, animal models involving the synthetic activation of maternal immune systems have highlighted alterations to internal brain circuitry that may support deleterious changes in offspring neurodevelopment underlying changes to behavior and cognition, with relevance for understanding the pathophysiology of human neurodevelopmental disorders [134, 190]. Exposure to inflammatory cytokines in maternal circulation provides a plausible mechanism for the link between MIA and altered offspring neurodevelopment [191‒194], though caution is warranted in interpretation where causation cannot be ascertained.
Critically, the majority of preclinical animal models of MIA have targeted rodent species, which may fail in translation to capture key alterations to highly derived neural circuitry implicated in the etiology of human neurodevelopmental disorders. Diverse brain systems nevertheless show unique patterns of altered offspring neurodevelopment subsequent to exposure to MIA in rodent models, with changes to frontostriatal [195‒197] and frontoamygdalar [198, 199] circuitry, including functional deficits in PV interneurons [200]. Ongoing research in the rhesus macaque model of MIA seeks to determine how maternal infection in gestation affects cytokine distribution and cellular changes in the nonhuman primate brain, including the expression and transcription of dopamine and inhibitory neurotransmitters in MIA-exposed nonhuman primate offspring.
Derived differences in the internal microstructure of the PFC, and its protracted course of development, may render the primate (and thus, human) brain uniquely vulnerable to the deleterious effects of maternal immune activation [201]. Longitudinal MRI study of a cohort of MIA-exposed macaques has shown reduced gray matter volumes in the frontal lobes and PFC in MIA-exposed offspring measured from 6 to 36 months of age [202]. Importantly, these alterations were correlated with impaired performance on reversal learning tasks, in a manner similar to impairments seen in schizophrenia and psychosis [203‒205]. A significant increase in extracellular free water, a potential biomarker for neuroinflammation seen in first-episode psychosis [206‒208], has also been noted in the MIA-exposed macaque [209].
At the cellular level, we have observed differences in dendritic morphology of pyramidal cells in the dorsolateral PFC in a pilot cohort of MIA-exposed macaque offspring [210]. Specifically, analyses of Golgi-stained neurons revealed reduced diameter as well as increased branching of apical dendrites in MIA-exposed offspring as compared to control animals. These results have been replicated in a subsequent cohort [211], which additionally found no significant differences between MIA-exposed offspring groups in aberrant dendritic development as a function of the timing of MIA exposure, either at the end of the first trimester or the end of the second trimester. Thus, exposure to MIA at different points in gestation can lead to similar effects on dendritic morphology and, by extension, the balance of inputs to cortical neurons.
Transcriptomic analyses have additionally found alterations to differential gene expression implicating deficits in synaptic development and myelination in the PFC [212], as well as the hippocampus in MIA-exposed primates. Further cell-specific analyses have shown alterations to cellular distribution, as well as genes associated with synaptic transmission, in region-specific patterns in the amygdala (Ander et al., in preparation [98]).
Interestingly, an altered pattern of increased striatal dopamine has also been observed in the MIA-exposed macaque [213]. Striatal dopamine is of particular interest in schizophrenia, where one of the most robust findings includes increased presynaptic dopamine levels [214, 215]. Further analyses of frontostriatal and frontoamygdalar organization and neurotransmitter systems may continue to reveal unique effects of MIA on highly derived neural circuitry that cannot be recapitulated in rodent models.
Inhibitory control of behavior develops in childhood and adolescence, guided by the maturation of the PFC and its connections [216, 217]. Human adolescence is marked by a dynamic period of changes in structural neuroanatomy, including myelination, as well as reorganization of local and systems-level connectivity [218]. Adolescence is also a uniquely vulnerable period for the etiology of psychiatric disorders and associated symptomatology, including schizophrenia and anxiety, owed in part to the refinement and maturation of GABAergic circuitry during this critical period [219‒221]. The course of onset and severity of deficits in inhibitory control in schizophrenia is highly variable, and a lack of longitudinal data limits our understanding of the development of executive dysfunction in the disorder [222]. By contrast, deficits in inhibitory control of behavior are common in ASD, emerge in early development, and are associated with the severity of restricted and repetitive behaviors [223‒225]. Disruptions to the balance between excitatory and inhibitory activity (E/I imbalance) may represent a common mechanism across disorders by which similarities in executive dysfunction arise in both ASD and schizophrenia, but divergences in both neurobiological and clinical phenotypes as well as symptomatology argue strongly for a cross-diagnostic, comparative approach [226]. Strikingly, both ASD and schizophrenia are associated with a significantly increased risk of developing dementia and early cognitive decline relative to NT peers [227, 228]. As E/I imbalance has been well documented in Alzheimer’s disease and mild cognitive impairment [229‒231], dysregulated E/I balance may contribute to accelerated cognitive decline and brain aging in ASD, schizophrenia, and other neurodevelopmental and psychiatric disorders.
Humans are characterized by the longest lifespan of any primate species, with an average pre-industrial lifespan roughly doubling since our divergence with our last common ancestor shared with chimpanzees and extending again as a function of cultural adaptations including access to better nutrition and medical care. Our striking longevity may uniquely predispose humans to deleterious conditions of aging, including cancers, heart disease, and neurodegeneration [232]. Cognitive decline differentially affects executive function, and altered performance on tasks designed to measure inhibitory behavioral control and behavioral flexibility may begin in early middle age [233]. Analysis of the processes that contribute to early cognitive decline in neurodevelopmental disorders such as E/I imbalance, neuroinflammation, and oxidative stress in neurodevelopmental disorders may provide key insights for understanding healthy aging in the human brain.
Conclusion
While subcortical structures in the human brain are highly conserved in their basic structures and functions, modifications of frontostriatal, frontoamygdalar, and GABAergic circuitry are important hallmarks of mammalian, primate, and human evolution. Changes to these systems subserve key adaptations in brain and cognition underlying changes to sensory processing systems, as well as highly complex social behavior. Many neurodevelopmental and psychiatric disorders are characterized by perturbations to inhibitory systems and circuitry, and subtle changes in cellular or synaptic distribution can contribute to altered balance of excitatory and inhibitory neuronal activity with widespread and diverse consequences for systemic brain function. Developmental processes and mechanisms, including maternal immune activation and increased vulnerability of inhibitory cells and synapses with age, may contribute to worsening symptomatology in neurodevelopmental disorders.
Acknowledgments
We would like to thank Profs. Daphne Soares and Muhammad Spocter, Dr. Grace Capshaw, and the speakers and participants of the JB Johnston Club for Evolutionary Neuroscience and Karger Workshop for their inspiring talks and conversation. We are grateful to the families of brain donors and to the staff and administration of the University of Maryland Brain Bank, Autism BrainNet, and the Williams Syndrome Association for their continued support of postmortem brain research. We would also like to thank Avni Shah for assistance in literature review as well as Melissa D. Bauman, Erin L. Carlson, Caroline Lew, Deion Cuevas, Kimberly Groeniger, and Lisa Stefanacci for their contributions to the research discussed herein.
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
Funding Sources
Funding for research has been provided in part by NIH grants R03MH103697 and R01MH097236.
Author Contributions
K.L.H. drafted the manuscript, D.M.Z.G. provided original data, C.M.S. and K.S. secured funding, and all authors contributed to literature review, writing, and editing of the manuscript.